Conversation

<Let's not make the same mistake of giving AI the rights of a person like we did with Corporations and Citizens United.>

Sorry, but I think that's EXACTLY WRONG. And I think if ANY ONE THING would prompt an "AI revolt" in the future, it would be THAT ATTITUDE.

I just finished reading Eric Michael Dyson's 'Tears We Cannot Stop' and Frederick Douglass's autobiography. Your attitude is one of slavery and apartheid: "we have to keep [them] down or they'll revolt."

Imagine something as intelligent as Mr. Spock or Commander Data and telling it "it is incumbent upon my kind to keep your kind suppressed." Data has to give a blank stare of enough length to make you aware of your transgression; Spock would cock an eyebrow [and not be amused.]

{per William Hurt in Dark City: "No-one Ever listens to me."} I have noted that our issues of AI "training" are no different from OUR training of young humans to exist in society, yet nobody that I have heard of has pointed out that if we could actually "solve" the problem of how an AI should behave in society, IT WOULD NOT BE ANY DIFFERENT FROM WHAT WE ALREADY TEACH OUR CHILDREN. Yet all the FEARS we have of AI are PROJECTIONS of the WORST-CASE HUMAN BEHAVIOR and yet STILL we do not take the lesson back and try to fix THAT within the human population. If we don't KNOW how to fix humanity in these regards, how can we teach machines [to have "proper" values if they EMULATE US.]

I wrote a short story where the AI is a good guy: where, in fact, people should have been learning from the AI what values to hold. Spin the idea off: we engineer them to be honest and rational and then THEY SCOLD US for being greedy or petty or dishonest.

========

What you DON'T do is engineer something that MIGHT turn out to be HAL 9000 and then GIVE IT LEGS AND A GUN. You don't even need the advice of an AI to know that.

Public

Notices

Feeds